My essay Predictable Competitors1 explored the assumption

Transcription

1 Predicting Competitors: Game Theory in This article outlines the results of a business war game designed to demonstrate the delusions that many pricers fall prey to when trying to predict their competitors next moves. By applying game theory to pricing, the author gives a unique perspective on the challenges faced by pricers on a daily basis. Author Mark Chussil is Founder and CEO of Advanced Competitive Strategies, Inc. and a Founder of Crisis Simulations International, LLC. A pioneer in business war-gaming and strategy simulation, his designs have won a patent (for CSI s DXMA simulator) and a best-inclass industry award (for ACS ValueWar ). He can be reached at mchussil@competing.com. My essay Predictable Competitors1 explored the assumption of predictability and the easy-to-fallinto traps of using competitors previous behavior to predict their future behavior. We also discussed how to avoid those traps and, in so doing, how to open up promising opportunities to make better strategy decisions. We did it in the context of pricing. Here we ll talk less about illusions of predictability and more about delusions of predicting. We ll do so in the context of a pricing tournament in which over 250 able strategists participated and for which I ran millions of simulations. The pricing tournament was a kind of massive business war game using humans strategies and a computer s calculations. Unless you already know what I m going to say if you think you can predict what I m going to say, write it down so you can check later you will have two surprises by the time this essay is over. The case Here s the situation I presented to those able strategists. You are a pricing strategist for a company with businesses in three industries. You will develop pricing strategies for each of those three businesses, covering 12 quarters (three years). In each industry your business has two competitors, and in each industry your business and your competitors businesses start from identical positions. You define success: you decide how much you care about profitability and market share. I proceeded to describe the three industries in some detail. Those details included these and more: The Ailing industry was shrinking, had somewhat price-sensitive customers, and was capital intensive. The Mature industry was growing slowly, had relatively priceinsensitive customers, and was labor intensive. The Fast Growth industry was growing rapidly, had price-sensitive customers, and was a bit on the capital-intensive side. Because the businesses in each industry began from identical positions, everyone had an equal opportunity to win. The only thing that would determine who won was the quality of their pricing-strategy decisions. Participants selected a pricing move for Q1 (the first of the 12 quarters), a pricing strategy for Q2-4, another pricing strategy for Q5-8 (year 2), and a third pricing strategy for Q9-12 (year 3). The three multi-quarter pricing strategies could be the same or different, in any combination. The Q1 pricing move would be cut, hold, or raise (see picture); the three subsequent pricing strategies would be selected from a list of strategy options. Just as in real life, participants had to select their strategies without knowing what their competitors would do. 1. Cut my price in Quarter 1 2. Hold my price in Quarter 1 3. Raise my price in Quarter 1 (I know what you re thinking. Rest assured, there are analytic nuances, technological marvels, and good answers for your good questions, none of which will we go into here. What we re about to cover doesn t depend on those nuances, marvels, and answered questions.) What would you do? Let s focus on the first move, the one affecting only Q1, where in each industry (Ailing, Mature, Fast Growth) you could cut, hold, or raise your price. What would you do in each industry? Write down your answer, or, if you re telling yourself you ll remember your answers, record them legibly in mental ink. 22 First Quarter 2010

2 Pricing Of course I haven t given you as much information as I gave the tournament participants. Still, though, you probably have some idea of what you d do, something along the lines of in a declining market it is best to prices or I d calculate the effect on profitability of my prices and then decide. Figure 1: Q1 Pricing-Strategy Decisions Second question. I didn t directly ask this one in the tournament but it s relevant for our discussion here. What do you think your competitors (i.e., the other strategists participating in the tournament) will do for the two competing businesses you ll face in each industry? Record those answers too. Notice how that question explicitly focuses your attention on competitors. An approach such as in a declining market it is best to prices addresses competitors only obliquely: you d probably consider the role of competition in a declining market, but perhaps not think about their specific actions. Now this third question: Did you think your competitors would do something different from what you chose to do? In my experience strategists tend to assume competitors will behave as they wish them to or as they have behaved in the past. Who knows, they might even assume competitors are not as clever, quick, or attentive as they are. Hence that third question. But why would you think your competitors will do something different from you in scenarios when we have explicitly said they start from positions identical to your own? Having now encountered that question, would you change your answer to the first question, the one about what you would do in each industry? Our first surprise Unless you correctly predicted me (did you?) and so there was no surprise, we ve completed our first surprise: the realization that we make assumptions unconsciously that we wouldn t make deliberately. (Remember, you heard it first here.) Further evidence appears as the subject of our second surprise. Our second surprise If the right pricing strategy were obvious, we would expect our able strategists to be pretty close to unanimous in their strategy selections for the tournament. They were not. In other words: surprise, we re all over the map on how to kick off an effective pricing strategy. Figure 2 shows the percentage of strategists who chose to cut, hold, or raise price in their Q1 pricing decision for each industry. Even the most-popular choice hold price in Q1 in the Mature industry was preferred by only 57% of the strategists. No move even got a majority in the Ailing and Fast Growth industries. In the Ailing industry, roughly equal percentages of strategists thought it would be best to cut or to raise their prices! I mentioned earlier that the strategists indicated their performance objectives: market share, profits, or any combination. Perhaps if we control for their objectives we ll see something closer to consensus. Here s as close as we get to consensus: Sixty-seven percent of the strategists who wanted a mix of share and profit in the Mature industry chose to hold their prices in Q1. No other pricing decision, in any of the industries, got anywhere near that level. For instance, Figure 3 shows the Ailing industry. The highest we see is 49% in favor of holding prices in Q1 to achieve a mix of share and profit, and 49% who d cut price to gain share. We do see an effect we might expect. The strategists who preferred First Quarter

3 Figure 2: Q1 Pricing Decisions (Mature) Figure 3: Q1 Pricing Decisions (Ailing) Figure 4: Q1 Pricing Decisions (Fast Growth) market share as their performance objective were more likely to cut price than to raise it, and those who sought profit were more likely to raise price than to cut. Even so, those effects seem muted: raising or cutting prices didn t win a clear majority of strategists. (That changed only when we looked at the most-extreme of the shareseekers, and we d have to get really extreme the 5 or 10 most sharehappy strategists, out of over 250 to approach a consensus.) Since objectives have a demonstrable effect on pricing decisions, we might argue that predicting competitors pricing moves translates, at least in part, to understanding their objectives. We would come to the same conclusions and the same surprises, though, because the strategists were far from unanimous about objectives too. There were many in the Ailing industry, for example, who wanted growth, and many in Fast Growth who wanted profits. We ll talk more about objectives presently. Let s flip the numbers around a bit. What are the odds that you would be wrong if you predicted your competitors would make the most-popular move? At best, you d have a 33% chance of being wrong, if you were in the Mature industry and knew your competitors wanted some share and some profit. At worst, you d have a 63% chance of being wrong for profit-oriented competitors in Fast Growth. (Figure 4. The odds of being wrong equal 100% minus the most-popular choice, which is 37% for hold. ) Not good odds. So far we ve focused on one quarter s pricing decisions. In the tournament our able strategists could choose from a longer list of strategies that ranged from aggressive to reactive, cooperative to confrontational, tend-to-raise to tend-to-cut, and so on, for their moves after Q1. There was nothing approaching consensus or even popularity in those decisions. 24 First Quarter 2010

4 Why might real life feel different? Here we ve seen that hundreds of strategists are far from agreement when they predict what pricing moves would work and, therefore, why it would be difficult to predict what any of them would do. So why might it seem that competitors are predictable in real life? Here are a few ideas. Although I don t have data to prove or disprove them, they are consistent with my experience working with thousands of strategists around the world, and consistent with the way that competitive-strategy tools think. (See Further Reading, below.) Clean slate. In the pricing tournament, the strategists started with clean slates: no history, no politics. No one could rely on, or had to defend, previous decisions. In real life, that s not the case. Safety. In the tournament, the worst that could happen was that a person wouldn t do terribly well in a simulation. Big deal. In real life, a person could lose his or her job for a change that seems to backfire. There s perceived safety in consistency: don t blame me, this strategy has worked well for [fill in suitable time span and/or credible other people]. In the tournament, no strategy has an inside track. Clarity. In the tournament, all the facts were laid out and the decisions were relatively simple. In real life, there s more complexity and ambiguity, which might make strategists wary of upsetting a precariously balanced system. No tradition. In the tournament, strategists were free to define success according to their own preferred combinations of market share and profit, unburdened by how we do things around here. In real life, different companies may assign similar missions to their businesses (e.g., fly full in airlines). We ve seen that definitions of success objectives influence pricing decisions. That effect is doubtless magnified by real-life industries oral traditions of what it takes to achieve objectives. In real life, competitors often emulate others moves. That happened in the tournament too, in various ways and for various reasons. However, that affected pricing after Q1, and here we focus mostly on the Q1 decisions. The phenomena after Q1 are fascinating but this essay may already be long enough to try your patience, and certainly mine. What our surprises mean Strategists face the problem of predicting competitors. We have just seen why doing so may be harder than we might have thought. Our first surprise suggests that we don t ask questions that might help us predict competitors (more on that quite soon). Our second surprise suggests that our competitors may not be so easy to predict, unless the slate isn t clean, jobs aren t safe, issues aren t clear, and tradition is binding. The predicting problem has two elements, carbon and silicon, a.k.a. humans and software. On the human side, we ve seen that we are prone to make optimistic, or at least unexamined, assumptions about what competitors will do. We know we assume, and we work in good faith not to do so. Take, for instance, SWOT (strengths, weaknesses, opportunities, and threats) analysis. There we endeavor to give equal time to our competitors so we don t get too convinced of our infallibility and invulnerability. But let us look at the SW of SWOT. Its strength is that it is easy, fast, portable, and potentially insightful. (I say potentially because whether we get insight depends as much on what we receive as on what it transmits.) Its weakness is well, let s illustrate by comparing SWOT to business war games. In a business war game you walk in your competitors shoes. In SWOT analysis you merely look at them. The predicting problem has two elements, carbon and silicon, a.k.a. humans and software. Ways to do better: Use competitive intelligence to learn about your competitors objectives. A change in objectives may well upset any pricing (or other) pattern you may have observed. Also use CI to learn about new management. Our able strategists have demonstrated that smart people differ in what they think will work. A change in management may foretell a change in strategy. That s true especially because now under new management rarely means still under previous thinking. Competitive dynamics resemble chess more than accounting or trend lines. Practice your game before the big real-life match. Business war games let you do that. In my experience, surprises like those we ve explored here are the rule, not the exception, in business war games. The good news is, it s a lot cheaper to get surprised during practice. During strategy debates, ask if you were our competitors, how would you take advantage of our move. Ask what could go wrong. Ask what are we assuming and do we believe what we are assuming. Then there s the software side. Our software thinks like us. (Who else would it think like?) We tell it how to think. We tell it that profit equals revenue minus costs. We tell it demand will change X% for a Y% change in price. We tell it how to combine those thoughts, and others, to figure out the bottom-line effects of price changes. Since our thinking includes assumptions, so does the thinking of the software that calculates on our behalf. Some of the able strategists, participating in the tournament at a pricing conference I addressed, actually wrote down their spreadsheet-style First Quarter

5 calculations on their strategy-decision forms: at this price, with these fixed and variable costs, here s how much I ll make. If such people were in their offices, it s likely they would run those calculations in Excel to guide their decisions. Those people chose a paradigm for how to decide before deciding on pricing strategies. That paradigm, like all others, has its paradigm-specific assumptions. In that case, the paradigm assumed competitors were simply irrelevant. Ways to do better: Ask your tools what assumptions they make. They re not talking? Okay, ask the tool-designers or -wielders what assumptions their tools make. Ask especially how competitors moves would be taken into account. Ask about competitive dynamics. What gets held constant over time, what doesn t. Ask about what-if. If there s anything we should take away from the Q1 pricing decisions we explored, let alone the similar variation in the other pricing decisions that we didn t explore, it s that we need to test the very real possibility that they will move in a surprising direction. (Subtle point: they may even want to move in a different direction, and may be watching us before they commit.) Worry more about the what-ifs than about precision. Who cares if the 63% chance of being wrong for the profit-oriented competitors in the Fast Growth industry should really be 62% or 64%? It s far more important to explore your chess opponent s options than to measure precisely where each piece is in its square. Bonus Surprise Did you predict there d be only two surprises since that s what I said at the beginning? Oh my poor student. There s one more surprise concerning the able strategists. We noted the variation in their pricing decisions. Every one of them believed that he or she was selecting the strategy that d win; if she or he believed otherwise, he or she would have selected a different strategy. Surprise: not all of them won. (I didn t either.) That means that we strategists don t know what will work. Actually, that s a slight overstatement. Some did know what would work, and their strategies performed well in the tournament, in which I ran over 25,000,000 (no joke) what-if simulations. The problem is, we don t know in advance whose strategies will work, nor do we know if they will be successful in the future. For what it s worth, and to add a minor surprise: the person who did best, so far, isn t a pricing expert. The person is a market-research practitioner. It s humbling and perhaps infuriating that we don t know what will work. We may assume solace in thinking that we do know what ll work in our industries, and those Ailing, Fast Growth, and Mature industries are pretty weird. Maybe that s true, but I don t think so. And I ll end by asking one more annoying question: if we re so good at pricing, where do price wars come from? Endnotes First Quarter 2010